LS-SVM based capacitor anomaly identification method

The sudden failures caused by ESR abnormal increase cannot be identified by the conventional aging failure criterion. A two-dimensional data-driven based method is proposed to identify the abnormal state of capacitors. First, the least square support vector machine (LS-SVM) algorithm, a modified version of the SVM algorithm, is proposed to identify the abnormal state of capacitors. The regularization parameters and relaxation variables of the LS-SVM algorithm are established by grid search method. Then, the proposed LS-SVM algorithm and conventional k-means algorithm are applied to identify the abnormal state of capacitors based on the accelerated aging test results. The results show that the accuracy of the proposed model is significantly improved compared with k-means algorithm, increasing the identification sensitivity of the state on the failure boundary.